Low-temperature behavior of disordered magnetic impurities: Distribution of effective Kondo temperatures

Several rare-earth compounds, low-dimensional organic conductors, and spin chains exhibit low-temperature divergences of their magnetic susceptibility and specific heat (non-Fermi-liquid behavior). Such divergences are often related to disordered ensembles of magnetic impurities in those systems....

Повний опис

Збережено в:
Бібліографічні деталі
Дата:2004
Автори: Zvyagin, A.A., Makarova, A.V.
Формат: Стаття
Мова:English
Опубліковано: Фізико-технічний інститут низьких температур ім. Б.І. Вєркіна НАН України 2004
Назва видання:Физика низких температур
Теми:
Онлайн доступ:http://dspace.nbuv.gov.ua/handle/123456789/119731
Теги: Додати тег
Немає тегів, Будьте першим, хто поставить тег для цього запису!
Назва журналу:Digital Library of Periodicals of National Academy of Sciences of Ukraine
Цитувати:Low-temperature behavior of disordered magnetic impurities: Distribution of effective Kondo temperatures / A.A. Zvyagin, A.V. Makarova // Физика низких температур. — 2004. — Т. 30, № 6. — С. 639-643. — Бібліогр.: 39 назв. — англ.

Репозитарії

Digital Library of Periodicals of National Academy of Sciences of Ukraine
id irk-123456789-119731
record_format dspace
spelling irk-123456789-1197312017-06-09T03:04:30Z Low-temperature behavior of disordered magnetic impurities: Distribution of effective Kondo temperatures Zvyagin, A.A. Makarova, A.V. Низкотемпеpатуpный магнетизм Several rare-earth compounds, low-dimensional organic conductors, and spin chains exhibit low-temperature divergences of their magnetic susceptibility and specific heat (non-Fermi-liquid behavior). Such divergences are often related to disordered ensembles of magnetic impurities in those systems. In this work the distribution function of the effective characteristic of a single magnetic impurity, the Kondo temperature, is derived. We calculate how distributions of Kondo temperatures depend on the effective dimensionality of the problem and on the concentration of impurities. 2004 Article Low-temperature behavior of disordered magnetic impurities: Distribution of effective Kondo temperatures / A.A. Zvyagin, A.V. Makarova // Физика низких температур. — 2004. — Т. 30, № 6. — С. 639-643. — Бібліогр.: 39 назв. — англ. 0132-6414 PACS: 75.20.Hr, 75.10.Pq, 71.10.Hf http://dspace.nbuv.gov.ua/handle/123456789/119731 en Физика низких температур Фізико-технічний інститут низьких температур ім. Б.І. Вєркіна НАН України
institution Digital Library of Periodicals of National Academy of Sciences of Ukraine
collection DSpace DC
language English
topic Низкотемпеpатуpный магнетизм
Низкотемпеpатуpный магнетизм
spellingShingle Низкотемпеpатуpный магнетизм
Низкотемпеpатуpный магнетизм
Zvyagin, A.A.
Makarova, A.V.
Low-temperature behavior of disordered magnetic impurities: Distribution of effective Kondo temperatures
Физика низких температур
description Several rare-earth compounds, low-dimensional organic conductors, and spin chains exhibit low-temperature divergences of their magnetic susceptibility and specific heat (non-Fermi-liquid behavior). Such divergences are often related to disordered ensembles of magnetic impurities in those systems. In this work the distribution function of the effective characteristic of a single magnetic impurity, the Kondo temperature, is derived. We calculate how distributions of Kondo temperatures depend on the effective dimensionality of the problem and on the concentration of impurities.
format Article
author Zvyagin, A.A.
Makarova, A.V.
author_facet Zvyagin, A.A.
Makarova, A.V.
author_sort Zvyagin, A.A.
title Low-temperature behavior of disordered magnetic impurities: Distribution of effective Kondo temperatures
title_short Low-temperature behavior of disordered magnetic impurities: Distribution of effective Kondo temperatures
title_full Low-temperature behavior of disordered magnetic impurities: Distribution of effective Kondo temperatures
title_fullStr Low-temperature behavior of disordered magnetic impurities: Distribution of effective Kondo temperatures
title_full_unstemmed Low-temperature behavior of disordered magnetic impurities: Distribution of effective Kondo temperatures
title_sort low-temperature behavior of disordered magnetic impurities: distribution of effective kondo temperatures
publisher Фізико-технічний інститут низьких температур ім. Б.І. Вєркіна НАН України
publishDate 2004
topic_facet Низкотемпеpатуpный магнетизм
url http://dspace.nbuv.gov.ua/handle/123456789/119731
citation_txt Low-temperature behavior of disordered magnetic impurities: Distribution of effective Kondo temperatures / A.A. Zvyagin, A.V. Makarova // Физика низких температур. — 2004. — Т. 30, № 6. — С. 639-643. — Бібліогр.: 39 назв. — англ.
series Физика низких температур
work_keys_str_mv AT zvyaginaa lowtemperaturebehaviorofdisorderedmagneticimpuritiesdistributionofeffectivekondotemperatures
AT makarovaav lowtemperaturebehaviorofdisorderedmagneticimpuritiesdistributionofeffectivekondotemperatures
first_indexed 2025-07-08T16:30:02Z
last_indexed 2025-07-08T16:30:02Z
_version_ 1837096976861102080
fulltext Fizika Nizkikh Temperatur, 2004, v. 30, No. 6, p. 639–643 Low-temperature behavior of disordered magnetic impurities: Distribution of effective Kondo temperatures A.A. Zvyagin1,2 and A.V. Makarova3,1 1B. Verkin Institute for Low Temperature Physics and Engineering of the National Academy of Science of Ukraine, 47 Lenin Ave., Kharkov 61103, Ukraine 2Max-Planck-Institut für Physik komplexer Systeme, Dresden, Germany E-mail: zvyagin@mpipks-dresden.mpg.de 3Kharkov State Economic University, Kharkov, Ukraine Received January 12, 2004 Several rare-earth compounds, low-dimensional organic conductors, and spin chains exhibit low-temperature divergences of their magnetic susceptibility and specific heat (non-Fermi-liquid behavior). Such divergences are often related to disordered ensembles of magnetic impurities in those systems. In this work the distribution function of the effective characteristic of a single mag- netic impurity, the Kondo temperature, is derived. We calculate how distributions of Kondo tem- peratures depend on the effective dimensionality of the problem and on the concentration of impu- rities. PACS: 75.20.Hr, 75.10.Pq, 71.10.Hf The Kondo effect [1] is the well-known example in which modern theoretical methods like renormali- zation group theory, the Bethe ansatz, bosonization, conformal field theory, etc. have demonstrated their strength [2,3]. It describes the effects of the exchange interaction between the spin of a magnetic impurity and spins of itinerant electrons. The crossover from the strong coupling to the weak coupling regime for a Kondo impurity manifests nonperturbative effects pre- sent in condensed matter theory. During the last decade the interest in non-Fer- mi-liquid behavior of magnetic systems and metallic alloys has grown considerably. A large class of con- ducting nonmagnetic materials does not behave as usual Fermi liquids (i.e., with the finite magnetic sus- ceptibility, finite value of the Sommerfeld low-tem- perature coefficient of the specific heat, and quadratic temperature behavior of the resistivity) at low tem- peratures. There are several possible reasons for such a behavior. One of the most known examples of the non-Fermi-liquid properties is the Kondo effect for multichannel (n is the number of channels) electron systems: For an impurity spin less than n/2 a non-Fer- mi-liquid critical behavior results [4]. Another reason, which can cause the non-Fermi-liquid behavior, is the presence of the quantum critical point [5] (i.e., a phase transition governed not by the temperature but by some other parameter, like pressure, chemical substitution, etc.). In that case fluctuations of the or- der parameter interact with itinerant electrons and can cause low-temperature divergences of thermody- namic characteristics. However, for most of nonstoichiometric conductors in which the non-Fermi-liquid behavior has been ob- served, see, e.g., the recent reviews [5–8], the mag- netic susceptibility and low-temperature specific heat usually manifest logarithmic or weak power-law be- havior with temperature, while their resistivity de- creases linearly (or with some power-law exponent less than 2) with temperature, showing a large resid- ual resistivity [9–19]. This is different from the pre- dictions of the theory of the overscreened Kondo effect [3,4]. For example, Ref. 14 reported the results of measurements of the magnetic susceptibility, NMR Knight shift, and low-temperature specific heat. To explain the observed features it was necessary to as- sume some disorder, with a distribution of Kondo tem- peratures of magnetic impurities. The inhomogeneous distribution of localized magnetic moments was later confirmed [18] by muon spin rotation experiments. © A.A. Zvyagin and A.V. Makarova, 2004 The above-mentioned properties and the alloy nature of the studied compounds suggest that the disorder (a random distribution of localized electrons or a random coupling to the conducting electron host) can play the main role in the low-temperature non-Fermi-liquid character of such systems. The idea of (unscreened) magnetic moments existing in disordered metallic sys- tems has been formulated in [20–24]. It was proposed that near the metal–insulator transitions (or for the sufficiently alloyed systems far from the quantum crit- ical point) disordered correlated conductors contain localized magnetic moments. The change in the inter- actions between the impurity sites and the host spins can be considered as a modification of the characteris- tic energy scale, the Kondo temperature TK . At that scale the behavior of the magnetic impurity manifests the crossover from the strong coupling regime (for T H TK, �� , H being the magnetic field) to the weak coupling regime T H TK �� , . The impurity spin be- haves asymptotically free in the weak coupling case, and it is screened by the spins of itinerant electrons in the strong coupling case. The random distribution of magnetic characteristics of impurities (i.e., their Kon- do temperatures) may be connected either with the randomness of exchange couplings of itinerant elec- trons with the local moments [21] or with the random- ness of the densities of conduction electron states [20]. Both types of randomness renormalize the single uni- versal parameter, the Kondo temperature, which char- acterizes the state of a magnetic impurity. A thorough comparison of experimental results for non-Fermi-liq- uid behavior of disordered heavy fermion Kondo al- loys has been performed with very good agreement with theoretical predictions of the model for distrib- uted Kondo temperatures [17]. Later it was pointed out that the problem of the behavior of magnetic im- purities with random distributions of their Kondo temperatures in metals can be solved exactly, with the help of the Bethe ansatz [24–26]. The role of the long-range (Ruderman—Kittel—Kasuya—Yosi- da) coupling between local moments was taken into account [25–28] (Griffiths phase [29] theory), exhib- iting properties qualitatively similar to those of mod- els with noninteracting local moments. Also, the pres- ence of a spin–orbit interaction in some disordered heavy fermion alloys demands the study of a magnetic anisotropy, which can play an essential role in physics of disordered spin interactions [26–28,30]. Finally, it was exactly proved that for correlated electrons sys- tems with magnetic impurities with random distribu- tions of their couplings to the host it is also possible to introduce a distribution of effective Kondo tempera- tures [31] which governs the low-temperature non- Fermi-liquid behavior of the system. It was pointed out [25,26,28,31] that distributions of effective Kondo temperatures for each magnetic im- purity can cause divergences of the magnetic suscepti- bility and the Sommerfeld coefficient of the specific heat for quasi-one-dimensional organic conductors and quantum spin chains, where such behavior was ob- served [32–35]. To explain power-law divergences of the magnetic susceptibilities and Sommerfeld coeffi- cients of rare-earth and actinide compounds, as well as quasi-one-dimensional organic conductors and quan- tum spin chains, it was necessary to use a distribution of Kondo temperatures (the strong-disorder distribu- tion, for which the tails are rather large) which starts with the term P T G TK K( ) ( )� � �� � 1 (� � 1) valid till some energy scale G for the lowest values of TK [25–27,30,31]. The goal of this work is to obtain the distribution of Kondo temperatures for a system with magnetic impurities and to show how such a distribu- tion will depend on the effective spatial dimensio- nality of the system (it turns out that the exponent � is different for three-dimensional non-Fermi-liquid heavy fermion systems and for quasi-one-dimensional organic conductors and quantum spin systems [16,32, 33,35]). Let us consider the model of electrons (itinerant and localized, where localization centers, i.e., 3d, 4f, or 5f orbitals, are distributed randomly on a hyper- cubic lattice with a random nearest-neighbor hopping) with the Hamiltonian [20] H H� � � � ��t j j c c j j j j( , ) , , , † , int , (1) where cj, † (cj, ) creates (annihilates) an electron with spin at site j, and t j j( , )� are hopping ele- ments. The interaction part of the Hamiltonian, H int � � �( ) , ,U/ n nj j j2 , is determined by the Cou- lomb interaction of localized electrons (the sum is over random positions of localized electrons, and n c cj j j, , † , � ). Generally speaking, one can add to the Hamiltonian (1) disordered local potentials � j j jn, . The hopping integrals can be approxi- mated [36] as overlap integrals t j j E r a r a j j j j ( , ) exp ... , , � � �� � � � � � � � � � � � � � � � 0 1 , (2) where a is the Bohr radius and E U0 ~ is the effective binding energy of the dopant (localized electron). We suppose that hopping integrals are random because of the disorder of the distribution of localized electrons. Let us consider the situation for which localized electrons are in the magnetic state, i.e., their valence is close to 1, which is satisfied if t j j EF 2( , ) ( )� ��� �� � j , U j� (with negative j , where �( )EF is 640 Fizika Nizkikh Temperatur, 2004, v. 30, No. 6 A.A. Zvyagin and A.V. Makarova the density of states at the Fermi level) [2]. It is natu- ral to assume that the density of the localized magnetic moments nl depends on the density of dopants n as n n n/nl � �exp( )max , (3) where nmax is related to the critical distance bet- ween localization centers Rc via n Vcmax /� 1 . Here Vc depends on the effective dimensionality of the distribution of localized magnetic moments. It is equal to V R /c c� 4 32� for the three-dimensional case, V Rc c� � 2 for the effectively two-dimensional situa- tion, and for effectively one-dimensional distribution of localized magnetic moments one has V Rc c� . Such a distribution is well-known in the theory of disor- dered systems (it follows from simple combinatorics), e.g., in the theory of dislocations with defects it is known as the Koehler formula [36], see also [37,38]. For large n, nl is a decreasing function of n, as it must be. Using such an assumption we can find the proba- bility P r( ) to find the nearest-neighbor localized mo- ment of a given site at a distance r, see below. (On the other hand, P r( ) can be derived independently [39], and we can obtain Eq. (3).) Obviously, the density of the localized magnetic moments can be written as n P r drl Rc � � � ( ) . (4) Using the definition of Rc we obtain P r n r nr / d r nr d nr ( ) exp( ), exp( ), exp( � � � � � � � � � � 2 3 2 4 3 3 2 2 ), d � � � �� � � � 1 (5) where d is the effective dimensionality of the distribu- tion of localized electrons, cf. [36,39]. One can check that nl has a maximum at nmax, which, actually, jus- tifies Eq. (3). The next step is to obtain the distribution of hop- ping integrals t j j( , )� between localized moments (in what follows we shall denote them simply as t). Equation (2) implies r t a E /t( ) ( )� ln 2 0 . Then, using ~( ) ( )( )P t P r dr/dt� and Eq. (5) we get ~( ) ln ( ) exp [ ( ) ln ( )], P t an t a E /t na / E /t d � � � �� �2 2 0 3 3 04 3 3 2 20 2 2 0 0 � �aln E /t na E /t d na E /t ( ) exp [ ( )], exp [ ( )] � � � ln ln , d � � � �� � � � 1 (6) for the three-, two-, and one-dimensional cases, respectively. The Kondo temperature of the localized magnetic moment can be written as [2] T U E t U U E t DK F F � � �� � � � � � � � 2 2 2 � � � ( ) exp | | ( ) ( ) exp | | ( ) ( ) � �� � � � � � U U E tF2 2 , (7) where we have assumed homogeneous distributions of the local potentials j � and introduced the low-energy cutoff D, as usual for the Kondo problem [2]. Notice that for � 0 one can define the Kondo temperature as T D U/ E tK F� �exp ( ( ) )2 2� . Then Eq. (7) implies dT dt t T T /D K K K � � 2 ln ( ) . (8) Defining x D/TK� ln ( ) and A U E E UF� �8 0 2� ( ) /| | ( ), we obtain the distributions of Kondo temperatures P T Z n n T x / Ax n/ n Ax K d K ( ) ( )ln ( ) exp [ ( ) ( max max � � 2 3 16 82 3ln )], ln( ) exp [ ( ) ( )], exp [ ( max ma d Ax n/ n Ax d n/ n � � � � 3 4 2 2 2ln x) ( )],ln Ax d � � � �� � � � 1 (9) for the three-, two-, and one-dimensional cases, re- spectively, where Zd are normalization constants. Observe that the divergence of P TK( ) as TK � 0 (due to the factor TK �1) is weakened by logarithmic factors like ln ( )T /DK . This is why P TK( ) can be normalized to unity over some interval 0 � �T TK max, where Tmax is given by T D U / U E EFmax exp ( | | ( ) ( ) )� � � �8 0 2 . Tmax can be related to the parameter G, see above. Obviously the distributions of Kondo temperatures depend on the parameters of impurity , the constant Low-temperature behavior of disordered magnetic impurities: Distribution of effective Kondo temperatures Fizika Nizkikh Temperatur, 2004, v. 30, No. 6 641 of Coulomb repulsionU, the density of states of itin- erant electrons at the Fermi level, the Bohr radius, and the density of localized orbitals. However, Eqs. (9) imply that such a set of parameters is realized in two main governing parameters: n/nmax and A. It turns out that for large enough intervals of val- ues of TK the distribution P TK( ) is proportional to TK ��1, i.e., it is of the form used in [25–27,30,31] without derivation, based on experimental results. The exponent � � 1 is determined by A and n/nmax. The plots of the logarithms of P TK( ) as functions of x are presented in Fig. 1,a for one-, two-, and three-di- mensional situations (we did not take into account the terms in Eqs. (9) which do not depend on x here; those terms yield only constant shifts). Here we used the parameters D, ,U, E0, appropriate for real disor- dered rare-earth and actinide systems, which produces A � 1. Also, according to experiments, one can expect n/nmax .� 0 01. Notice that the maximal values of intervals of TK , Tmax, depend on A via Tmax � � �D /Aexp ( )1 , and, hence, x /Amin � 1 . This is why we present here the results for large enough values of x (i.e., for small TK), that the distributions in the considered intervals are reminiscent of the power laws � �TK 1 �. One can see that the exponents of the distri- butions depend on the effective dimensionality of the distribution of the localized electrons. To illustrate how the effective exponents depend on A and on the concentration of localized electrons n, we present results for the logarithms of the distribu- tions of Kondo temperatures for A � 5, n/nmax .� 0 01 in Fig. 1,b and for A � 1, n/nmax .� 0 5 in Fig. 1,c. The results show that the effective exponents � are large ( .� � 0 8 for the realistic case A ~ 1, n/nmax .~ 0 01 for three space dimensions). It is too large compared to the experiments on three-dimensional heavy fermion systems [14,16–18], for which values of � � 0.6–0.85 were observed. Also, for some values of the parameters a value � � 1 was obtained, which implies the absence of low-temperature divergences in such cases. The rea- son for such an overestimation can lie in the disorder in the distribution of local potentials of localizated electrons, j , which was not taken into account in our description. However, smaller values of � (0.26–0.42) have been observed for low-dimensional organic con- ductors and spin chains [32,33]. It turns out that large values of � ~ .0 2 were observed for disordered ensem- bles of impurities close to a metal-insulator transition. Such a distribution of Kondo temperatures pro- duces non-Fermi-liquid behavior of many characteris- tics of the system [25–27,31]. For example, the ground state magnetization of the system behaves as M H G� ( / )� (H is an external magnetic field), i.e., essentially in a nonlinear way. The static magnetic susceptibility at low temperatures is divergent as T /G� ��1 . Also, the nonlinear behavior of the electron specific heat c T/G~ ( )� implies divergent behavior of the Sommerfeld coefficient. The low-temperature be- havior of the correlation length is proportional to ( )G/T �, which implies non-Fermi-liquid behavior of the low-temperature resistivity. Finally, the imagi- nary part of the dynamic magnetic susceptibility is also divergent as G T g T� � �� �2 1 ( / ), where g x( ) � � � � ��x dy/y x y 1 1 2 2� ( ) is a universal scaling func- tion [25]. Summarizing, in this work we have calculated the distribution of effective Kondo temperatures for a 642 Fizika Nizkikh Temperatur, 2004, v. 30, No. 6 A.A. Zvyagin and A.V. Makarova 2 4 6 8 3 4 5 6 7 8 9 10x A = 1, n/n = 0.01max 2 3 4 5 6 7 8 9 x43 5 6 7 8 9 10 A = 5, n/n = 0.01max 2 4 6 8 3 4 5 6 7 8 9 10x A = 1, n/n = 0.5max a b c P (T ) K Fig. 1. Logarithms of the distributions of the Kondo temperatures for the three-dimensional (solid line), two-dimensional (dotted line), and one-dimensional (dashed line) cases. system of correlated electrons in which itinerant elec- trons are hybridized with randomly distributed local- ized electrons (3d, 4f, or 5f orbitals). Those distribu- tions depend on many parameters: concentration of localized orbitals, their Bohr radii, Coulomb interac- tion of localized electrons on orbitals, effective ener- gies of localized electrons, bandwidths of itinerant electrons, etc. However, we have shown that only two parameters effectively govern the behavior of the dis- tribution functions of Kondo temperatures. One of those parameters is the ratio of the density of localized electrons to the maximal density (in a hypercubic lat- tice). The other parameter is related to the maximal Kondo temperatures for which such an approach can be applied. We have shown that distributions of Kondo temperatures depend on the effective dimen- sionality of the disordered localized electrons. Due to such a disordered behavior of magnetic impurities many thermodynamic and kinetic characteristics of the considered systems manifest non-Fermi-liquid features in their low-energy characteristics. 1. J. Kondo, in: Solid State Physics: Advances in Re- search and Applications, F. Seitz, D. Turnbull, and H. Ehrenreich (eds.), Academic, NY (1969), V. 23, P. 184. 2. A.M. Tsvelick and P.B. Wiegmann, Adv. Phys. 32, 453 (1983). 3. I. Affleck, Acta Phys. Polon. 26, 1869 (1995). 4. P. Schlottmann and P.D. Sacramento, Adv. Phys. 42, 641 (1993). 5. F. Steglich, J. Magn. Magn. Matter. 226, 1 (2001). 6. H. v. Löhneysen, F. Huster, S. Mock, A. Neubert, T. Pietrus, M. Sieck, O. Stockert, and M.M. Waffen- schmidt, Physica B230–232, 550 (1997). 7. M.B. Maple, M.C. de Andrade, J. Herrmann, Y. Da- lichaouch, D.A. Gajewski, C.L. Seaman, R. Chau, R. Movshovich, M.C. Aronson, and R. Osborn, J. Low Temp. Phys. 99, 223 (1995). 8. G.R. Stewart, Rev. Mod. Phys. 73, 797 (2001). 9. C. Seaman, M.B. Maple, B.W. Lee, S. Ghamaty, M.S. Torikachvili, J.-S. Kang, L.Z. Liu, J.W. Allen, and D.L. Cox, Phys. Rev. Lett. 67, 2882 (1991). 10. G.R. Stewart, Phys. Rev. B47, 3208 (1993). 11. B. Andraka, Phys. Rev. B49, 348 (1994). 12. B. Andraka, Phys. Rev. B49, 3589 (1994). 13. M.C. Aronson, R. Osborn, R.A. Robinson, J.W. Lynn, R. Chau, S.L. Seaman, and M.B. Maple, Phys. Rev. Lett. 75, 725 (1995). 14. O.O. Bernal, D.E. MacLaughlin, H.G. Lukefahr, and B. Andraka, Phys. Rev. Lett. 75, 2023 (1995). 15. K. Matsuhira, T. Sakakibara, and H. Amitsuka, Phy- sica B206–207, 326 (1995). 16. M.C. de Andrade, R. Chau, R.P. Dickey, N.R. Dilley, E.J. Freeman, D.A. Gajewski, M.B. Maple, R. Mov- shovich, A.H. Castro Neto, G.E. Castilla, and B.A. Jones, Phys. Rev. Lett. 81, 5620 (1998). 17. C.H. Booth, D.E. MacLaughlin, R.H. Heffner, R. Chau, M.B. Maple, and G.H. Kwei, Phys. Rev. Lett. 81, 3960 (1998). 18. D.E. MacLaughlin, R.H. Heffener, G.J. Nieuwenhuys, G.M. Luke, Y. Fudamoto, Y.J. Uemura, R. Chau, M.B. Maple, and B. Andraka, Phys. Rev. B58, R11849 (1998). 19. L. Shlyk, J.C. Waerenborgh, P. Estrela, L.E. De Long, A. de Visser, and M. Almeida, J. Phys.: Con- densed Matter 11, 3525 (1999). 20. R.N. Bhatt and D.S. Fisher, Phys. Rev. Lett. 68, 3072 (1992). 21. V. Dobrosavljeviæ, T.R. Kirkpatrick, and G. Kotliar, Phys. Rev. Lett. 69, 1113 (1992). 22. E. Miranda, V. Dobrosavljeviæ, and G. Kotliar, Phys. Rev. Lett. 78, 290 (1997). 23. E. Miranda, V. Dobrosavljeviæ, and G. Kotliar, J. Phys.: Condensed Matter 8, 9871 (1996). 24. A. Klümper and A.A. Zvyagin, Phys. Rev. Lett. 81, 4975 (1998). 25. A.A. Zvyagin, Phys. Rev. B62, R6069 (2000). 26. A.A. Zvyagin, Fiz. Nizk. Temp. 28, 1274 (2002) [Low Temp. Phys. 28, 907 (2002)]. 27. A.H. Castro Neto, G. Castilla, and B.A. Jones, Phys. Rev. Lett. 81, 3531 (1998). 28. A. Klümper and A.A. Zvyagin, J. Phys.: Condensed Matter 12, 8705 (2000). 29. R.B. Griffiths, Phys. Rev. Lett. 23, 17 (1969). 30. A.H. Castro Neto and B.A. Jones, Phys. Rev. B62, 14975 (2000). 31. A.A. Zvyagin, Phys. Rev. B63, 033101 (2001). 32. L.N. Bulaevsky, A.V. Zvarykina, Yu. S. Karimov, R.B. Lyubosky, and I.F. Shchegolev, Zh. Eksp. Teor. Fiz. 62, 725 (1972) [JETP 35, 384 (1972)]. 33. K. Ikegami, S. Kuroda, M. Saito, K. Saito, M. Sugi, T. Nakamura, M. Matsumoto, and Y. Kawabata, Phys. Rev. B35, 3667 (1987). 34. K. Mukai, K. Suzuki, K. Ohara, J.B. Jamali, and N. Achiwa, J. Phys. Soc. Jpn. 68, 3078 (1999). 35. S.V. Demishev, R.V. Bunting, L.I. Leonyuk, E.D. Obraztsova, A.A. Pronin, N.E. Sluchanko, N.A. Sama- rin, and S.V. Terekhov, Pis’ma v Zh. Eksp. Teor. Fiz. 73, 36 (2001) [JETP Lett. 73, 31 (2001)]. 36. J.S. Koehler, in: Imperfections in Nearly Perfect Crystals, New York (1952), P. 197. 37. C.L. Bauer, Philos. Mag. 11, 827 (1965). 38. A. Granato, and K. Lücke, J. Appl. Phys. 27, 583 (1956). 39. M. Lakner, H. v. Löhneysen, A. Langenfeld, and P. Wölfle, Phys. Rev. B50, 17064 (1994). Low-temperature behavior of disordered magnetic impurities: Distribution of effective Kondo temperatures Fizika Nizkikh Temperatur, 2004, v. 30, No. 6 643